On 04/12/18 07:51, Tejun Heo wrote:
On Wed, Apr 11, 2018 at 07:58:52PM -0600, Bart Van Assche wrote:
Several block drivers call alloc_disk() followed by put_disk() if
something fails before device_add_disk() is called without calling
blk_cleanup_queue(). Make sure that also for this scenario a request
queue is dissociated from the cgroup controller. This patch avoids
that loading the parport_pc, paride and pf drivers triggers the
following kernel crash:
BUG: KASAN: null-ptr-deref in pi_init+0x42e/0x580 [paride]
Read of size 4 at addr 0000000000000008 by task modprobe/744
Call Trace:
dump_stack+0x9a/0xeb
kasan_report+0x139/0x350
pi_init+0x42e/0x580 [paride]
pf_init+0x2bb/0x1000 [pf]
do_one_initcall+0x8e/0x405
do_init_module+0xd9/0x2f2
load_module+0x3ab4/0x4700
SYSC_finit_module+0x176/0x1a0
do_syscall_64+0xee/0x2b0
entry_SYSCALL_64_after_hwframe+0x42/0xb7
Reported-by: Alexandru Moise <00moses.alexander00@xxxxxxxxx>
Fixes: a063057d7c73 ("block: Fix a race between request queue removal and the block cgroup controller")
Signed-off-by: Bart Van Assche <bart.vanassche@xxxxxxx>
Cc: Alexandru Moise <00moses.alexander00@xxxxxxxxx>
Cc: Tejun Heo <tj@xxxxxxxxxx>
So, this might be correct for this reported case but I don't think
it's correct in general. There's no synchronization between blkcg
q->lock usages and blk_cleanup_queue(). e.g. hotunplugging and
shutting down a device while file operations are still in progress can
easily blow this up.
Hello Tejun,
I have retested hotunplugging by rerunning the srp-test software. It
seems like you overlooked that this patch does not remove the
blkcg_exit_queue() call from blk_cleanup_queue()? If a device is
hotunplugged it is up to the block driver to call blk_cleanup_queue().
And blk_cleanup_queue() will call blkcg_exit_queue().
Bart.